Monday, September 9, 2019

Decarboxylative etherification

Our most recent work in electrochemistry was published today: A method to make hindered ether compounds from abundant carboxylic acids. Given the extensive information in the >400 pages of SI, we are not going through the details of the work in this blog post, but instead decided to present some details from behind the scenes.

As a project that required massive work, you might be wondering how it got started. Actually, Phil always got the consulting question that how to construct hindered ether bonds effectively from the pharmaceutical industry because they are quite important in drug discovery. It surprised us that in the 21st century, people still rely on acid promoted hydroalkoxylation to make such bonds, which often fails in the real-world cases due to a lack of chemoselectivity and sluggish reactivity. Being aware of this, we decided to tackle this problem.

By digging into the literature, we found an interesting reaction called the Hofer-Moest reaction which has a >100-year old history. This reaction enables generation of carbocations that can be intercepted by solvent quantities of alcohol to form the ether products. Without even understanding the details in the original paper published in 1902 (written in German), we started our journey towards the hindered ether synthesis. Carefully studying the follow up literature, we found this reaction is extremely limited in terms of the scope of both carboxylic acids and alcohols. In order to develop a practical method that can be used by the industry, we thought the key was to find a suitable solvent and electrolyte that enabled us avoiding use solvent amount of alcohol. After extensive evaluation (>1000 experiments), we identified a new set of conditions for this old chemistry. To make a long story short, here is an example of just how simplifying this chemistry is, in the context of synthesis of biologically active intermediates. Using compound 1 which is an intermediate of aurora kinase modulator as an example, previously, a multi-step route was employed requiring over 6 days of reaction time (<4% overall yield), wherein the key C–O bond-forming reaction—the treatment of methylenecyclobutane with BF3•Et2O—provides the ether in only 11% yield. Now with our electrochemical method, it can be accessed in only 2 steps, 51% yield and most impressively 15h.

While we think the step-count and time savings largely speak for themselves, our colleagues who are industrial process chemists are much more rigorous about efficiency characterization.

Process chemists tend to use “PMI,” or Process Mass Intensity, as a quantitative measure of a synthetic route’s efficiency. Our friends at BMS have conveniently developed an app, the “PMI Prediction Calculator” (Click here to play with the app yourself), that enables people quickly to know the efficiency and PMI of a certain process. We used this app to calculate the PMI of the old vs. new route to 1. The result: Our route to synthesis of compound 1 has a PMI of 252, versus 1304 for the previous one. This result clearly demonstrates the high efficiency, robustness and greenness of our new method.

Ideality is another way to quantify the parameters of efficiency for a synthetic sequence. We calculated the ideality% for the 12 real-word applications based on the definition published a few years ago (this paper). Amazingly, for 9 examples, the current methods have an ideality of 100% which are much higher than the reported routes.

Lastly, it is a tradition for our lab to showcase the limitations of the methodology that we have developed, to give the readers a better understanding of the reaction. Some of the limitations for the decarboxylative etherification and hydroxylation are shown below. In general, the secondary acids without any stabilizing effect such as benzyl or hetero atom will be problematic for this etherification reaction, due to the relatively instability of the corresponding carbocations. We hope this may be helpful for those who decide to try this reaction on similar substrates.

Let us know if you have any questions or comments regarding the work! Thanks for reading!

Ming, Jinbao and the etherification team


  1. Very nice, concept-opening work. Forgive me few comments: 1) This is very useful because it generates carbocations under neutral conditions, and Lwis acid promoted addition to olefins tends to be quite dirty. But please have you compared your method with carbocations generated by thermolysis of triflates also? (Tertiary alcohol, Tf2O, 2,6-di-t-Bu pyridine 2 equivs, -50C, then add alcohol and warm up? I have seen conditions like this for O-benzylation of hindered secondary alcohol with BnOTf generated in situ from BnOH and Tf2O, it worked in suprisingly good yield.

    2) Using over-stoechiometric silver salt slightly detracts from the process friendliness on scale, for now. Molecular sieves (sodium aluminosilicate and sodium-potassium aluminosilicate) are known to exchange cations (they are very good at gobbling up Cs from organic solution, for example). Have you tried some other water scavenger (maybe CaO) that does not have this liability to absorb silver cation?

    3) I noticed you are using Bu4NPF6 as electrolyte salt but AgSbF6 as silver salt (but there is AgPF6 mentioned in the paper). Is it for practical consideration like AgSBF6 solubility in DCM? (I remember working with AgSbF6 and it was very well behaved). Bu4NPF6 is quite a grease-ball. I know that smaller quaternary ammoniums like Me4N(+) and Et4N(+) are quite toxic, but have you tried something like quaternized DABCO cations? Those are well behaved, crystalline, polar (so it is easy to remove on silica), and resistant towards oxidation.

    Thank you. Excellent study.

    1. Thanks for your insightful comments and questions! See itemized responses below.

      1. We know that there are some special cases that can be achieved by using the traditional SN1 chemistry as you mentioned. However, they are far from general in terms of yield and functional group compatibility. Therefore, we aimed to provide a more general solution to hindered ether synthesis by using non-acid and non-basic conditions which is completely different from any traditional chemical method. We haven’t directly compared the conditions that you mentioned (maybe in an upcoming issue of chemTweetChem?) but we think it is likely that the chemical conditions won't be general if you want to make secondary-tertiary ethers or tertiary-tertiary ethers, based on the extensive search of literature. It's also important to note retrosynthetic differences in those two methods -- for triflation/SN1, you need the tertiary alcohol, which in our view is somewhat less ubiquitous than the many commercial alpha-tertiary acids that the published reaction takes advantage of.

      2. Using 1.5 eq of Ag salt could is definitely a drawback of this reaction. So we have tried to remove the Ag salt and it turns out that for many substrates, we can use KSbF6 instead without affecting the yield significantly. We are not completely sure about the role of Ag salt at this stage, but we observed Ag plating at the cathode after the reaction and the voltage of the whole reaction decreases lot when using Ag as the additive. Agreed that there may be room for further optimization here.

      3. It’s a good idea to try CaO as the water scavenger; we haven’t tested that. However, we don’t think absorption of Ag cations by molecular sieves is significant, as the reaction yield is almost identical even when we use 3 or 5 equiv of Ag salt.

      For non-activated tertiary carboxylic acids, we observed better yields when using AgSbF6 instead of AgPF6, possibly because of the solubility as you mentioned. Indeed, we have tried a series of quaternary ammoniums such as Me4NCl, Et4NBr, TBAClO4, TBAOTs. While most of the quaternary ammoniums gave inferior results than TBAPF6, TBAClO4 was able to give a slightly better yield if it was used together with AgClO4. However, we chose TBAPF6 as the final electrolyte because of the potential explosive issues of perchlorates, especially for large scale reactions. We haven’t tried DABCO salts before; probably it’s another good choice for the electrolyte, but we like the commercial availability of TBA salts.


    2. Thank you, for a very thorough and informative reply

  2. also, one possible role for Ag(+) might be formation of Ag(2+) that brings about oxidation of radical to carbocation, which can happen on the anode surface even without silver also. The silver plating should happen on the cathode, so this would be the way Ag is lost, by reduction.

  3. An interesting way. Isn't there an additional reaction with group -CO2Me?

    1. Thanks for your inquiry! The ester group was able to survive in our reaction as demonstrated by the example 29 in the paper. The ester has to be hydrolyzed to free carboxylic acid before engaging into the electro-decarboxylative process.